How to process images of a video, frame by frame, in video streaming using OpenCV and Python

Question

I am a beginner in OpenCV. I want to do some image processing on the frames of a video which is being uploaded to my server. I just want to read the available frames and write them in a directory. Then, wait for the other part of the video to be uploaded and write the frames to the directory. And , I should wait for each frame to be completely uploaded then write it to a file.

Can you tell me how can I do it with OpenCV (Python)?

Edit 1: I wrote this code for capturing the video from a file, while new data are being appended at the end of the file. In other words, the out.mp4 file is not a complete video and another program is writing new frames on it. What I'm going to do is, wait for the other program to write new frames then read them and display them.

Here is my code:

import cv2
cap = cv2.VideoCapture("./out.mp4")

while True:
    if cap.grab():
        flag, frame = cap.retrieve()
        if not flag:
            continue
        else:
            cv2.imshow('video', frame)
    if cv2.waitKey(10) == 27:
        break

So the problem is the cap.grab() call! When there is no frame, it will return False! And it won't capture frames anymore, even if I wait for a long time.

Solution 1

After reading the documentation of VideoCapture. I figured out that you can tell VideoCapture, which frame to process next time we call VideoCapture.read() (or VideoCapture.grab()).

The problem is that when you want to read() a frame which is not ready, the VideoCapture object stuck on that frame and never proceed. So you have to force it to start again from the previous frame.

Here is the code

import cv2

cap = cv2.VideoCapture("./out.mp4")
while not cap.isOpened():
    cap = cv2.VideoCapture("./out.mp4")
    cv2.waitKey(1000)
    print "Wait for the header"

pos_frame = cap.get(cv2.cv.CV_CAP_PROP_POS_FRAMES)
while True:
    flag, frame = cap.read()
    if flag:
        # The frame is ready and already captured
        cv2.imshow('video', frame)
        pos_frame = cap.get(cv2.cv.CV_CAP_PROP_POS_FRAMES)
        print str(pos_frame)+" frames"
    else:
        # The next frame is not ready, so we try to read it again
        cap.set(cv2.cv.CV_CAP_PROP_POS_FRAMES, pos_frame-1)
        print "frame is not ready"
        # It is better to wait for a while for the next frame to be ready
        cv2.waitKey(1000)

    if cv2.waitKey(10) == 27:
        break
    if cap.get(cv2.cv.CV_CAP_PROP_POS_FRAMES) == cap.get(cv2.cv.CV_CAP_PROP_FRAME_COUNT):
        # If the number of captured frames is equal to the total number of frames,
        # we stop
        break

Solution 2

Use this:

import cv2
cap = cv2.VideoCapture('path to video file')
count = 0
while cap.isOpened():
    ret,frame = cap.read()
    cv2.imshow('window-name', frame)
    cv2.imwrite("frame%d.jpg" % count, frame)
    count = count + 1
    if cv2.waitKey(10) & 0xFF == ord('q'):
        break

cap.release()
cv2.destroyAllWindows() # destroy all opened windows

Solution 3

According to the latest updates for OpenCV 3.0 and higher, you need to change the Property Identifiers as follows in the code by Mehran:

cv2.cv.CV_CAP_PROP_POS_FRAMES

to

cv2.CAP_PROP_POS_FRAMES

and same applies to cv2.CAP_PROP_POS_FRAME_COUNT.

Hope it helps.

Solution 4

In openCV's documentation there is an example for getting video frame by frame. It is written in c++ but it is very easy to port the example to python - you can search for each fumction documentation to see how to call them in python.

#include "opencv2/opencv.hpp"

using namespace cv;

int main(int, char**)
{
    VideoCapture cap(0); // open the default camera
    if(!cap.isOpened())  // check if we succeeded
        return -1;

    Mat edges;
    namedWindow("edges",1);
    for(;;)
    {
        Mat frame;
        cap >> frame; // get a new frame from camera
        cvtColor(frame, edges, CV_BGR2GRAY);
        GaussianBlur(edges, edges, Size(7,7), 1.5, 1.5);
        Canny(edges, edges, 0, 30, 3);
        imshow("edges", edges);
        if(waitKey(30) >= 0) break;
    }
    // the camera will be deinitialized automatically in VideoCapture destructor
    return 0;
}

Solution 5

The only solution I have found is not to set the index to a previous frame and wait (then OpenCV stops reading frames, anyway), but to initialize the capture one more time. So, it looks like this:

cap = cv2.VideoCapture(camera_url)
while True:
    ret, frame = cap.read()

    if not ret:
        cap = cv.VideoCapture(camera_url)
        continue

    # do your processing here

And it works perfectly!

View more solutions

Comments

[Mehran]

Thank you very much for your answer. I did this in python. But the problem is, when you call cap>>frame; it doesn't wait for the frame to completely ready! So if the frame is not ready, then it will return null. And after that, it doesn't capture any other frames. Does it make sense?

[user2239318]

@Mehran would you be so gentle to share the python version?

[Gulzar]

This works, but at some random point the frames stop comin alhough the input stream continues. Something still gets hanged here

[JohnAllen]

This doesn't answer the question.

[Richard]

Including a link to further information would have been appreciated.

[Captain Fantastic]

I believe your last line should be: cv2.destroyAllWindows()

[Mehran]

Nice solution. But, what's the overhead of reinitializing the Capture? we need to be sure that destroying the object and creating it doesn't drop the performance.

[Alexander Lysenko]

It doesn't drop the performance - it drops few frames. There is no lugging or something similar, so this worked for us.

[Mehran]

This solution works when the stream (in this case, Camera) always starts from the current frame. In my case, the file is being uploaded to the server and saved to out.mp4. So if I reinitialize the Video capture, it will start from the first frame of the video not the last available frame.

[Rohan Pillai]

For OpenCV 3.0 onwards, change cap.get(cv2.cv.CV_CAP_PROP_POS_FRAMES) to cap.get(cv2.CAP_PROP_POS_FRAMES).

[Bahae El Hmimdi]

you must add if not ret: continue

[Batselot]

What is your count doing here?

[Bahae El Hmimdi]

counting number of frames proceeded.

[Batselot]

Oh okay, you weren't using them(printing or returning them) so I thought maybe you were doing something else with the count.